Transcript Determinants of FPI over FDI

Liquidity, Institutional Quality,
and the Composition of
International Equity Flows
Itay Goldstein (Univ of Pennsylvania)
Assaf Razin (Tel Aviv and Cornell Univ)
Hui Tong (International Monetary Fund)
Conference on International Adjustment,
Brussels; 9-10 November 2007
Motivation
 Little is known about the determinants of equity flows
(Foreign Direct Investment vs Foreign Portfolio
Investment.) from source countries.
 Traditionally, multinational firms engage in FDI, while
collective investment funds in FPI. But as in 2006 World
Investment Report, investment funds have become
important sources of FDI. Their cross-border M&As,
reached $135 billion and accounted for 20% of global
M&As.
 More work is needed to understand the composition of
equity flows.
2
Key Questions and Approaches
 This paper focuses on the impact of liquidity on
investors' choice between FDI and FPI.
 We first extend the theoretical model of Goldstein
and Razin (2005) by introducing aggregated liquidity
shock and capital market transparency. The
extended model predicts that countries with higher
probability of aggregate liquidity crises will be the
source of more FPI and less FDI.
3
Findings
 To test the prediction, we examine the variation of FPI
v.s. FDI for 140 source countries from 1985 to 2004.
 Our key explanatory variable is the estimated probability
of aggregate liquidity shocks, as proxied by net sale of
foreign assets (international reserves, credit and equity) .
 The probability of liquidity crises indeed has strong
impacts on the composition of foreign investment, as
predicted by our model.
 Moreover, greater capital market opacity in the source
country strengthens the impact of the crisis probability.
4
Implication of Our Findings
 Our findings help explaining the recent trend of collective
investment funds as growing sources of FDI, and shed
some light on how this trend may be affected by liquidity
crunch.
5
Outline




Theoretical Model
Data
Empirical Model and results
Conclusion
6
FPI vs FDI
 A key difference between FDI and FPI: FDI investors
have the management of the firms under their control;
but FPI investors delegate decisions to managers.
 Hence, direct investors are more informed than portfolio
investors regarding projects, which enables them to
manage projects more efficiently.
 But this info advantage comes with a cost. If investors
need to sell the projects before maturity due to liquidity
shocks, the price they can get will be lower when buyers
know that they have more info on the project.
 Tradeoff between management efficiency and liquidity.
7
Production Function
Three periods: 0, 1, 2; Project is initially sold in Period
0 and matures in Period 2.
1
R  K (1   )  AK 2
2
Production function
cdf  G ( ), G (1)  0, G ( 1)  1, g ( )  G '( )
Distribution
Function
8
Value of FDI till Maturity
In Period 1, after the realization of the productivity
shock, the direct Investor observes the shock and
chooses the level of K:
K * ( ) 
Expected Return
at t=0
1 
A
 (1   )(1   ) 1  1    2  E (1   ) 2

E
 A



A
2
A
2A




9
Value of FPI till Maturity
 Portfolio Investor has no information on the productivity
parameter, and will instruct the manager to maximize the
expected return. The chosen level of K then is:
K
Expected return
at t=0
1
A
 (1   ) 1  1
E


2A  2A
 A
10
Liquidity Shocks and
Resale of Investment
 Before maturity, investors may need to sell the project
due to liquidity shock;
 Some direct investors may also pretend that they are
facing liquidity shock (after they observe a bad
realization of the fundamental);
 Period-1 resale price is the asset’s expected value from
the buyer's viewpoint.
11
Resale Value of FDI
threshold     D
Productivity level under which the direct owner is selling
with no liquidity shock.
D
P1,D 
(1  ) 
1
1
(1  )2
g ( )d     1 2 g ( )d 
2A
2A
1
(1  )G( D )  
D
Resale value from Bayes’ law
(Lamda: the exogenous likelihood of liquidity shock)
2
(1


)
D
P1,D 
2A
The owner sets the threshold so that she is indifferent
between the price paid by buyer and the return if
continuing to hold the asset.
12
Resale Value of FPI
 If a portfolio Investor sells the asset, everybody knows
that it does so only because of the liquidity shock. Hence
the resale value of FPI:
1
1  2
1
P1, P  
g ( )d  
2A
2A
1
Since
 D  0  P1, D  P1, P
13
Aggregate Liquidity Crises
 Here we assume liquidity shocks to individual investors
are triggered by aggregate liquidity crises.
 During crises, some investors have deeper pockets than
others, and thus are less exposed to the liquidity issues.
Constrained direct investors need to sell, but they will get
a low price, in that buyers do not know whether the sale
is due to adverse info or liquidity constraint.
 Hence, the attractiveness of FDI decreases, when the
probability of a liquidity crisis becomes higher.
14
The Role of Opacity
 The effect of liquidity shocks on FPI/FDI is driven by lack
of transparency about the fundamentals of the direct
investment.
 If the fundamentals of each direct investment were
publicly known, then liquidity shocks would not be that
costly for direct investors, as the investors would be able
to sell the project at fair price without bearing the
consequences of the lemons problem.
15
Data
 Stocks of FPI and FDI in market value, for source
countries, from Lane and Milesi-Ferretti (2006). It has the
data on foreign assets for 140 source countries for the
past 30 years. We use the sample from 1985 to 2004,
due to higher data availability and reliability.
 The convention for distinguishing between portfolio and
direct investment is whether the ownership of shares of
companies is above or below the 10 percent threshold.
16
Empirical Model
 Panel Model
ln  FPI / FDI i ,t   ln  FPI / FDI i ,t 1   X i ,t 
 Pr  Liquidity Shock i ,t 1  ui  it
 Control variables: GDP, GDP per capita, stock
market capitalization, trade openness, lagged
real exchange rate appreciation, and year
dummies.
 The probability of aggregate liquidity shock at
next period is estimated from the Probit model.
17
Liquidity Crisis
 We define liquidity crisis as episodes of negative
purchase of external asset, based on the IFS Balance of
Payments dataset.
 13% of the sample from 1985 to 2004 experienced
liquidity crisis. Developed economies, such as Denmark,
Japan, New Zealand and Spain, also experienced crises.
18
Table 2: Episodes of Sales of External Assets
Argentina
2001,1989,1987,1986,
Brazil
1999,1997,1986,
Denmark
1994,
Greece
2001,2000,1997,1995,1992,1989,
Hong Kong S.A.R. of China
2001,1998,
India
1995,1990,1989,1988,1987,1986,
Indonesia
2001,
Israel
1988,1987,
Japan
1999,
Mexico
2002,2000,1994,1992,1988,
New Zealand
1997,1992,1991,1988,
Philippines
2001,2000,1997,1990,1987,
Spain
1994,
Turkey
2001,1994,
19
Probit Estimation
 We include explanatory variables that are unlikely to
affect the FDI and FPI composition directly, but will affect
the probability of crisis.
 They include: the US real interest rate, source country
current account balance/GDP, source country political
risk or Standard and Poor’s rating for source countries.
20
Table 3: Probit Estimation of Liquidity Crises
Case 1
Case 2
GDP, log
GDP per capita, log
-0.10***
-0.17***
-0.051
-0.082
US real interest rate
0.11***
0.10**
Current Account Balance/GDP
-0.012*
-0.072***
Government Balance/GDP
-0.010
-0.015
Trade openness
-0.18*
-0.072
Political risk
S&P country rating
Financial risk
-0.012**
0.00095
-0.24***
0.029
Observations
1658
736
R-squared
0.10
0.16
21
Determinants of FPI and FDI
 Higher probability of liquidity crises is associated with
higher log(FPI stock/FDI stock). See Table 4.
22
Table 4: Determinants of FPI over FDI
Fixed Effect
Dynamic
Prob of Liquidity Crisis at t+1
GDP, log
GDP per capita, log
Stock market capitalization
Trade openness
Real exchange rate (lag)
Ln(FPI/FDI), lag
4.49***
-3.29***
-0.44
0.19***
-0.86***
-1.57***
3.08***
-1.71
0.16
0.063
-0.39*
-0.61***
0.70***
Observations
Number of countries
738
60
656
57
23
Determinants of FPI and FDI
-Levels
 As a robustness check, we find that higher probability of
crisis is negatively associated with FDI stock, while
positively associated with FPI stock (Table 5).
 Hence, what we find is more than the pure capital flight
effect. Some may argue that during crisis, it is relatively
easier to fly money out of the country in the form of
portfolio investment than in the form of direct investment.
The capital flight argument would imply that FPI and FDI
rise together during crisis, which, however, is
inconsistent with Table 5.
24
Table 5: Determinants of FPI over FDI-level
FDILevel
FDILevel
Fixed
Effect
Dynamic Fixed
Effect
Dynamic
Prob of Liquidity Crisis at t+1
-0.25
-1.82***
2.07*
1.08
GDP, log
1.46***
-0.48
-2.27***
-0.52
GDP per capita, log
1.96***
0.49***
1.34***
0.12
Stock market capitalization
0.054
0.079*** 0.30***
0.066**
Trade openness
0.99***
0.022
0.088
0.0042
Real exchange rate (lag)
0.74***
0.038
-0.29
-0.23*
Log of FDI, (lag)
Log of FPI, (lag)
FPILevel
FPILevel
0.67***
0.71***
25
Proxies of Capital Market Opacity
 Pricewaterhouse-Cooper opacity index (OPA) in year 2001.
Including: corruption, efficacy of the legal system, deleterious
economic policy, inadequate accounting practices, and
detrimental regulatory structures.
 Pricewaterhouse sub index on accounting (ACC).
 Disclosure score from the 1995 Center for International Financial
Analysis and Research Report (CIFAR), measured by the
number of items disclosed in annual reports.
 The financial disclosure index in the 1999 Global
Competitiveness Report (GCR), measuring the perceptions of
company CEOs about a country.
26
Country
Acc
Country
Acc
Country
Acc
Finland
17
Argentina
30
Taiwan
40
Belgium
17
India
30
Brazil
40
Germany
17
Venezuela
30
Poland
40
USA
20
UK
33
Russia
40
Canada
20
Denmark
33
Egypt
40
Chile
20
Hong Kong
33
Czech Rep
44
Israel
20
Australia
33
Turkey
44
Thailand
20
Austria
33
Lebanon
44
Japan
22
S. Africa
33
Singapore
50
Indonesia
22
France
33
Spain
50
Sweden
25
Mexico
33
Portugal
50
Switzerland
25
Pakistan
33
Hungary
50
Ecuador
25
Saudi Arabia
33
Greece
50
Colombia
29
Philippines
33
China
56
Malaysia
30
Netherlands
38
Korea
30
Ireland
38
27
Results on Opacity
 Higher opacity increases the effect of the predicted
liquidity crisis on the FPI/FDI composition.
 Again, the findings support the existence of the liquidity
consideration. The pure capital flight effect will not
assign a role to the interaction of the source country
corporate transparency and the probability of crisis.
{Note that when transparency is included in the Probit
estimation, it does not turn out to be significant there}.
28
Table 6: Determinants of FPI stock over FDI stock-Opacity
OPA
OPA
ACC
Fixed
Effect
Dynamic
-5.25
Prob of Liquidity -3.55
Crisis
Prob(crisis)*
Opacity(OPA)
Prob(crisis)*
Opacity(ACC)
Prob(crisis)*
Opacity(CIFAR)
Prob(crisis)*
Opacity(GCR)
0.20*
ACC
CIFAR
CIFAR
GCR
GCR
Fixed DyEffect namic
Fixed
Effect
Dynamic
Fixed DyEffect namic
0.45
-0.81
-16.7**
1.86
-2.10
-0.94
0.15*
0.10*
0.50***
0.012
1.61
1.07
0.17**
29
Robustness Check-Probit
 In the Probit estimation, we substitute the ICRG political
risk indexes with Standard and Poor’s short-term
sovereign rating. The results are in Table 7.
 The effects of transparency become more significant.
30
Table 7: Determinants of FPI over FDI-Opacity
(Alternative Estimate of Crisis Probability)
OPA
OPA
ACC
ACC
CIFAR
CIFAR
GCR
GCR
Fixed
Effect
Dynamic
Fixed
Effect
Dynamic
Fixed
Effect
Dynamic
Fixed
Effect
Dynamic
Prob of Crisis
-15.3**
-6.93** -11.9*** -4.45**
-8.34**
2.00
-17.8***
-8.94**
Prob( crisis)
*Opacity(OPA)
0.36**
0.15**
0.23**
-0.087
5.28***
2.65**
Prob( crisis)
*Opacity(ACC)
Prob( crisis)*
Opacity(CIFAR)
Prob(crisis)
*Opacity(GCR)
0.36***
0.12**
31
Robustness CheckActual Occurrence of Crisis
 Evidently, this may create endogeneity issues in
estimation. But it still serves as useful checks,
particularly if there is some concern about the
forecasting power of Probit models.
 The dynamic panel estimation results are presented in
Table 8, with four proxies of opacity (OPA, ACC, CIFAR,
and GCR). Again, there we find that the occurrence of
liquidity crises at t + 1 increases the ratio of FPI to FDI.
Moreover, the impact becomes larger for source
countries with opaque capital markets.
32
Table 8: Determinants of FPI over FDI-Transparency
(Actual Occurrence of Liquidity Crisis)
OPA
ACC
CIFAR GCR
Liquidity Crisis (t+1)
-0.59**
0.12
-0.41** -0.51
Liquidity Crisis (t+1)
*Opacity(OPA)
0.014**
Liquidity Crisis (t+1)
*Opacity(ACC)
Liquidity Crisis (t+1)
*Opacity(CIFAR)
Liquidity Crisis (t+1)
*Opacity(GCR)
-0.0038
0.014**
0.19*
33
Conclusion
 Countries that have a high probability of
aggregate liquidity crises will be the source of
more FPI and less FDI. Moreover, the effect will
be more pronounced when corporate
transparency declines.
34